Sir,I read with great interest the article by Parashar et al.[1] on prevalence and correlates of nicotine dependence among construction site workers in Delhi. The authors had done a commendable job, but I have few issues in its methodology and serious limitations. First, the study methodology used convenient sampling which may not be reliable to estimate the prevalence at population level.[2] Convenient sampling severely undermines not only the external validity, i.e. generalizability, but also the internal validity of any study.[3] Further, those who did not give consent to participate and hence didn’t participate in the study among the construction workers at the site, their profile including socio-demographic variables and nicotine dependence behaviour may be entirely different from those who participated in the study leading to selection bias. A comparison of at least sociodemographic variables of these two groups would have eliminated this bias.[2] The high prevalence of tobacco users (91%) in the current study could be due to this effect as it is a common bias seen in cross-sectional studies with low response rate as seen in the current study which has an overall response rate of 68.8% only. The authors mentioned in the limitations section that the study is at least generalizable to sampled population, but as selection of patients was done nonrandomly through convenient sampling with low overall response rate, its generalizability even to the sampled population of construction workers is doubtful.These points severely undermine the validity of the findings of the study.